Help to choose PC Based Control Software + add ons.

J

Dear All, I am presently designing a Semiconductor Processing System which I intend to run using a PC Based Control scenario. It is a very high quality expensive product($3M+) so cost is not really a consideration. I am very familiar with PLC product/design/systems/interfaces, etc. I am not however familiar with PC based controls although from my initial evaluations to date, Think & Do plus Wonderware's Factory Suite 2000 seem to be the highest used products. My project requires Alarm/Auto-Dial-up/High Quality GUI's/Server /Database/Security/multipile Interfaces even PLC's/etc/etc. It must dial out to remote servers on an hourly basis. It must provide real-time data to a number of key managers in a factory situation/ Process Mgr /Equipment Mgr/ Production Mgr, etc. on separate PC's (dumb). It must allow video capabilities. It must communicate especially well with PC Anywhere, Microsoft Products, VB/C++ GUI's+HMI's, etc. It is also likely that parts of the system design will need to be changed every couple of years and I would like the software to be designed in sections so the section which is new only needs new software. Note:It must look and feel as a single system. All GUI's/HMI's will be developed in C++ and will sit between the user and the PC based control application. I will be hiring 3 Engineers for the Software part alone but I am still trying to get my budgets as close as possible at this stage. Also I am interested in info on the latest hardware/HMI panels/Computers /I/O's/etc,etc. Regards, John Quinn. CEO. NuEon Systems. P Paul Tolsma > I am presently designing a Semiconductor Processing System which I intend to run using a PC Based Control scenario. It is a very high quality expensive product($3M+) so cost is not really a consideration.<

Wow... must be nice!

> I am very familiar with PLC product/design/systems/interfaces, etc. I am not however familiar with PC based controls although from my initial evaluations to date, Think & Do plus Wonderware's Factory Suite 2000 seem to be the highest used products.<

PC control sounds like fun for this, assuming that you can stand any reliability issues associated with using the PC as your control
platform vs. a PLC. If the PC crashes, how much product do you lose and are there any safety issues if the controller... isn't? There are at least a half dozen packages on the market today for PC control. Steeplechase is the one that comes readily to mind, altho Intellution, GE Fanuc, AB also have PC based control software. ABs
is interesting in that it is their ladder logic... but it runs on the PC, not a PLC CPU.

> My project requires Alarm/Auto-Dial-up/High Quality GUI's/Server/Database/Security/multiple Interfaces even PLC's/etc/etc. It must dial out to remote servers on an hourly basis. It must provide real-time data to a number of key managers in a factory situation/ Process Mgr /Equipment Mgr/ Production Mgr, etc. on separate PC's (dumb).<

I would think this would all be in the HMI or SCADA. It needen't have anything to do with the controller. If that's the case, then Wonder or
FIX or your favorite package running over the control platform would be fine. Is there a reason to do it in the controller?

> It must allow video capabilities.
> It must communicate especially well with PC Anywhere, Microsoft Products, VB/C++ GUI's+HMI's, etc.<

Um... what kind of video capabilities? Presumably you'd like something like COGNEX to share its info directly with the controller memory rather than via physical I/O? We are thinking that way too, altho we haven't gotten to trying it yet. PC Any won't care what is under the OS... if the remote control capability works with whatever
software mix you wind up with, then any of that software will be available for use from the remote platform. As far as interacting with the machine and process from somewhere else in the plant (or on the planet) that again is a standard option for most HMI packages. Something like PC Any usually only comes in when you want to change
logic, not run the machine.

> It is also likely that parts of the system design will need to be changed every couple of years and I would like the software to be designed in sections so the section which is new only needs new software. Note:It must look and feel as a single system.<

Hire good engineers and develop the architecture at least twice before programming any of it <g>. Good thing there is a fat budget.

> All GUI's/HMI's will be developed in C++ and will sit between the user and the PC based control application.<

You mean you want to write your own? Why? It can certainly be done... but the future maintenance and support will drive the users berserk.

> I will be hiring 3 Engineers for the Software part alone but I am still trying to get my budgets as close as possible at this stage.<

If price is no object, the software will be the least of it. Most of the present PC-based controls packages seem to be priced between
5k and 10k. This is in the noise after yourself, three engineers, and the system hardware <g>... or, at least, the delta between packages is.

> Also I am interested in info on the latest hardware/HMI panels/Computers /I/O's/etc,etc.<

Umm... yeah. What type of I/O do you envision for this system? Board level in the PC backplane? Be careful with that... my last experience four years
ago was that board level stuff is not as well conditioned or robust as PLC I/O, altho once we stopped tripping over the little things it worked
really well. Stand alone I/O like OPTO or Wago? Or one of the major's I/O like AB or Siemens without the CPU? How big a system is this anyway in terms of physical I/O?

Sounds like fun... can you give us a little more background?

Paul T

T

Tanweer Ahmed

Hi Quinn,

Though I didn't personally used a PC based control system but recently attended a seminar on it. My understanding is that there is little difference in a PC based and a PLC based control system. In a PC based control system the CPU of the PC performs the function of that of a PLC CPU as well. This means that the same PC is used as PLC CPU as well as the HMI computer. Whereas the I/Os, communications and other functions
remains almost similar. The major disadvantage of the PC based control system noted is the
reliability of the PC. It is advisable to use a PC in industrial chasis preferably with redundant processor and hard disk. All the major PLC manufacturers are coming up with PC based control system.

Also visit http://softplc.com

If my above understanding is not accurate I would like to correct myself.

Regards,

Tanweer Ahmed
[email protected]

M

Michael Griffin

At 14:17 07/01/00 -0500, Tanweer Ahmed wrote:
<clip>
>The major disadvantage of the PC based control system noted is the reliability of the PC. It is advisable to use a PC in industrial chasis preferably with redundant processor and hard disk.
<clip><
I have been looking into this issue lately for my own purposes, and have come to the following conclusions.
The items on a PC which seem to fail most are hard drives, power supplies, and monitors. The fans which are found on newer CPUs are also a
potential source of failure, but I haven't noticed any problems with these yet (the hardware which has been around for a while didn't have these in them).

If reliability is an important factor, you should try take the above into account. Monitors are simple for your maintenance department to diagnose and replace provided the connectors are easily accessable.
Hard drives and power supplies are a different matter, as most standard PCs are not designed to be easily serviced on a plant floor.
However, there are industrial chassis which are available which allow easy replacement of the power supplies (plug in without opening the case). You should see if you can get a case which provides power supply test points or voltage diagnostic LEDs. You can also get inexpensive adaptors which allow the hard drives to plug into the normal mounting hole from outside the case.

The problem I haven't solved to my satisfaction yet is how to configure redundant hard drives when using Windows NT. I have received a lot of conflicting information on this, and I would really like to see this
demonstrated to me to answer the question definitively.
As for redundant processors, this is a much more difficult problem. Redundant processors is not the same thing as dual processors. I have not seen a standard PC which offers true processor redundancy. Since I have not had a CPU failure on our plant PC hardware, but have had far too many hard drive and power supply failures, I would worry about them and forget the
much more difficult (and expensive) CPU issue.

**********************
Michael Griffin
[email protected]

R

Don't leave out Beckhoff TwinCAT (Total Windows Control Automation Technology, www.beckhoff.com). It's been around for years, is fully IEC 1131-3 compliant, with a bunch of motion capabilities. It only works on NT
and they write their own real-time kernal so no dependency on third-party software.

It has been used for years in Europe and is just now making inroads in the US. Quite impressive stuff!

Wilmington, NC USA

R

Rick L. Hudson EMCO Inc.

Check out www.vmic.com for the IOWorks PC based control pkg

RH

W

Woodard, Ken CLEV

Robert Trask writes"PC control sounds like fun for this, assuming that you can stand any reliability issues associated with using the PC as
your control platform vs. a PLC. If the PC crashes, how much product do you lose and are there any safety issues if the controller... isn't?"

PC control, properly done, has reliability at the same level as the PLC, and maybe even greater, it all depends on the operating system that is responsible for the crashes, and not so much the hardware! At OLIN we have operated direct PC control in an electrochemical manufacturing operation in
Brazil for over 10 years with standard NEC 386-16 desktop computers, which by the way were upgraded during the 4th quarter to avoid Y2K issues, not
because the hardware was not doing it's job! A sister plant in the US also got new computers and software, at maybe a fraction of the cost of ten years ago but continues to utilize the same I/O system, can't do that with a PLC. PC control is much more adapatble to hot shadowing back up than the PLC, making it a much more robust platform than the PLC for mission critical operations. Direct PC control is being practiced on large scale mission critical applications within our company for well over a decade. In addition
it is being done utilizing multiple control nodes on ethernet networking, with deterministic control between nodes.

C

Curt Wuollet

AFAIK The Items that fail the most on PC's are Operating Systems and programs
by perhaps two orders of magnitude. If you look at it in that light, commodity
hardware looks real good. I'll bet you have more downtime due to software than
hardware, it's just easier to fix.

Curt Wuollet
Wide Open Technologies

Michael Griffin wrote:
>
> At 14:17 07/01/00 -0500, Tanweer Ahmed wrote:
> <clip>
> >The major disadvantage of the PC based control system noted is the
> >reliability of the PC. It is advisable to use a PC in industrial chasis
> >preferably with redundant processor and hard disk.
> <clip>
> I have been looking into this issue lately for my own purposes, and
> have come to the following conclusions.
> The items on a PC which seem to fail most are hard drives, power
> supplies, and monitors. The fans which are found on newer CPUs are also a
> potential source of failure, but I haven't noticed any problems with these
> yet (the hardware which has been around for a while didn't have these in them).
>
> If reliability is an important factor, you should try take the above
> into account. Monitors are simple for your maintenance department to
> diagnose and replace provided the connectors are easily accessable.
> Hard drives and power supplies are a different matter, as most
> standard PCs are not designed to be easily serviced on a plant floor.
> However, there are industrial chassis which are available which allow easy
> replacement of the power supplies (plug in without opening the case). You
> should see if you can get a case which provides power supply test points or
> voltage diagnostic LEDs. You can also get inexpensive adaptors which allow
> the hard drives to plug into the normal mounting hole from outside the case.
>
> The problem I haven't solved to my satisfaction yet is how to
> configure redundant hard drives when using Windows NT. I have received a lot
> of conflicting information on this, and I would really like to see this
> demonstrated to me to answer the question definitively.
> As for redundant processors, this is a much more difficult problem.
> Redundant processors is not the same thing as dual processors. I have not
> seen a standard PC which offers true processor redundancy. Since I have not
> had a CPU failure on our plant PC hardware, but have had far too many hard
> drive and power supply failures, I would worry about them and forget the
> much more difficult (and expensive) CPU issue.

C

Curt Wuollet

AFAIK The Items that fail the most on PC's are Operating Systems and programs by perhaps two orders of magnitude. If you look at it in that light, commodity hardware looks real good. I'll bet you have more downtime due to software than
hardware, it's just easier to fix.

Curt Wuollet
Wide Open Technologies

Michael Griffin wrote:
>
> At 14:17 07/01/00 -0500, Tanweer Ahmed wrote:
> <clip>
> >The major disadvantage of the PC based control system noted is the reliability of the PC. It is advisable to use a PC in industrial chasis
preferably with redundant processor and hard disk.
> <clip>< <
>
> I have been looking into this issue lately for my own purposes, and
> have come to the following conclusions. ...
<clip><

M

Michael Griffin

At 18:51 10/01/00 +0000, Curt Wuollet wrote:
>AFAIK The Items that fail the most on PC's are Operating Systems and programs
>by perhaps two orders of magnitude. If you look at it in that light, commodity
>hardware looks real good. I'll bet you have more downtime due to software than
>hardware, it's just easier to fix.
<clip>

>Earlier, I wrote:
<clip>
>> The items on a PC which seem to fail most are hard drives, power
>> supplies, and monitors. The fans which are found on newer CPUs are also a
>> potential source of failure, but I haven't noticed any problems with these
>> yet (the hardware which has been around for a while didn't have these in
>> them).
<clip>

My latest rely:

Perhaps in a general sense, but this is not the case for any of the test equipment I deal with. The following is rather lengthy, but I would
like to give you details of three typical examples:

Equipment 'A' (approx 10 installed - the oldest is 6 years)
- PC hardware is name brand 486 PCs.
- Operating sytem is DOS 6.22.
- Application program is approximately 12,000 lines of 'C' (developed with LabWindows for DOS) running under a DOS protected mode extender.
- Data aquisition hardware is several National Instruments boards.
- Software problems - none ever that I know of for DOS or the application (I make occasional inquiries since I wrote the original software and am looking for any feedback on my work). I would interpret this as meaning that any software problems occur too seldom for people to remember.
- Hardware problems - several per year involving power supplies or hard drives. I don't count monitor failures as they are easy to replace and don't cause down time for this equipment (the equipment runs 24 hours a day,
7 days a week on automatic lines).

Equipment 'B' (approx 4 installed, but 2 are brand new)
This is series produced equipment purchased from a company which specialises in this test area and has been in the business for much longer than PCs have been around.
- PC hardware is rack mount, but using standard PC components. The newer ones allow hard drives to be swapped more easily.
- Operating system is Windows 95.
- Data aquisition hardware is custom.
- Application software is the standard version written by the test equipment builder (of course).
- Software problems - not significant provided no changes have been made to the equipment (e.g. problems occur when installing a new version).
Better hardware diagnostics would be nice though.
- Hardware problems - at least two hard drive failures. The oldest machine is approximately 3 years old.

Equipment 'C' (one installed - approx 18 months old)
- Desktop style PC clone Pentium.
- Operating system is Windows NT
- Data aquisition is by GPIB
- Application software is custom written for this single application. It was written in Visual Basic by the small test equipment
specialist company which assembled the entire system.
- Software problems - A few minor ones, mainly associated with new versions which haven't been debugged properly. Windows NT will occasionally require rebooting. No significant total downtime has occurred though, except
if new software that hasn't been debugged yet was installed.
- Hardware failures - The monitor failed and had to be replaced.
- The computer itself was replaced recently for reasons not associatd with hardware failure. This is perhaps too small a hardware sample to be representative.

I have other examples, but the above gives a good sample of the cases on which I based my earlier comments. All of the above hardware is
exposed to a similar environment. The plant is relatively clean, and all are in free standing enclosures (mainly Rittal PC enclosures) with filtered ventilation and are not exposed to any vibration. Hardware failures do not seem to be correlated to the time of year, so hot weather does not seem to be a cause. There is no reason to suspect problems with the electrical service (e.g. voltage spikes, etc.). A power supply or hard drive failure will typically result in several hours of down time.

It is interesting to note that of all the PC based equipment we have (including ones not mentioned above), the ones which use DOS seem to have the least software or OS problems, while the ones which use Windows NT seem to have the most. Windows 95 seems to fall somewhere in between. This would seem to be contrary to the theoretical arguments which are made in Windows NT's favour (and against Windows 95 or DOS).

Example 'A' involves equipment for which we would like to implement a new, much improved test method later this year or early next. This will involve writing completely new software. We would also like to replace the computers with new ones while we are at it. The operating system will be an up-to-date one as well, since this equipment will be expected to have a long
subsequent life. From the comments listed above you can see that the PC related problem I will be trying to solve will be hardware failures, not
software failures. Hardware problems have cost us lost production, software problems have not in this case.

It is worth noting though that with examples 'B' and 'C' the companies which built them commented on how difficult it was to develop
their test software to run on a computer using a Windows NT or 95 operating system. They had started their projects with high hopes assuming they had no need for special real time considerations, and found out that Windows
introduced timing uncertainties much larger than they had expected as well as other mysterious problems. Close timing control of the test hardware was critical to meeting cycle time targets.
They also found it very difficult or impossible to obtain useful technical information on Windows to assist them in solving their particular problems. The typical Windows reference materials which are available are
not oriented towards this particular area of application. This may perhaps contribute to the sort of application software bugs you may be experiencing, as it is more difficult to make software reliable by design if you have to
determine operating system characteristics by conducting your own tests.
Testing may not reveal the complete range of responses from the OS.
It is unfortunate, but Windows seems to inject an element of mystery (and sometimes misery) in every machine in which it is used. Having questioned whether using desk top PCs was such a good idea after all, I am now beginning to wonder whether Windows NT is such a great idea either for this particular application.

I once had a conversation with one of our electricians to the effect that I thought we ought to stop buying a certain brand of equipment, as we seemed to have a lot of problems with it. He remarked that he thought we should stick with that brand, as he had gotten rather good at fixing them. I replied that I would be much more impressed with machines with which he
never had any reason to become familiar with - and when he thought about that, he had to admit that I was right.

**********************
Michael Griffin
[email protected]
**********************

J

James Bouchard

Some manufactures make server duty PC's that come with a wide variety of redundancy and hot swappable components and monitoring packages. We are presently using Compaq for this and the IS people swear by them not at them. They do cost more and our is people put them in the computer room which is clean and cool but you could just as well put them in appropriate enclosures
on the plant floor if you really need to.

James Bouchard

J

Johnson Lukose

<clip>
> It is interesting to note that of all the PC based equipment we
have
> (including ones not mentioned above), the ones which use DOS seem to have
> the least software or OS problems, while the ones which use Windows NT
seem
> to have the most. Windows 95 seems to fall somewhere in between. This
would
> seem to be contrary to the theoretical arguments which are made in Windows
> NT's favour (and against Windows 95 or DOS).
<clip>

Must be the reason that W95 is known as DOS7.0!

thanks.

N